1 Abstract

The goal of this project was to check if age, religion, gender and previous voting preferences has an effect on how “you” vote in Brexit Referendum Vote. Multi-Group CFA was used to test out this theory. After applying those methods on the given data set, we could not come to a conclusion of singling out a particular question variable which might explain the brexit referendum voting outcome.

2 Introduction

People of a country were always divided by their political ideas. The division was quite small until a few years ago. To explain in a graphical format. If neural ideas are a center line, the division in question is the difference in current political right to current political left. See the image below for reference (Ghosh 2019). Pundits warn, if the division continues to grow, it might cause a physical divide in the population. This trend seems to be occurring across the globe (Thomas Carothers 2019).

The reason for choosing Multi-Group Confirmatory Factor Analysis (MGCFA) is simple, in our lecture we learnt that invariance = equivalence. MGCFA satisfies the conditions mentioned in the lecture. Items act the same across the group. Factor structure is same across groups. Paths are equal across the groups. Latent means are almost equal across the groups.

This project aims to if previous vote, sex, religion and age group give variance in the Brexit voting preference. The central idea is to debunk the theory of “natural” alignment to a political party. Individual of a certain Race/Ethnicity/Age will align with a political party.

2.1 About the data

The date is provided by Dr. Erin M. Buchanan, Harrisburg University of Science and Technology and it’s explanation from a study website by Bob Altemeyer, it be found here (Altemeyer and Johnson 2020). The data is a scientific study of where do “you” lie on the Right Wing Authoritarians (RWA) scale (Altemeyer 2006)

Summary of the data after dropping unused columns.

library(lavaan)

## This is lavaan 0.6-7

## lavaan is BETA software! Please report any bugs.

library(semPlot)
library(knitr)

scoring = function(model){
  round(fitmeasures(model, c("chisq", "df", "cfi", "rmsea", "srmr")),3)
}

master = read.csv("study4.csv", stringsAsFactors = T)
master = master[, c(2,8,9,12,15,16,17,18,19,39:60)]
summary(master)

##        id          Remove        country       ageGroup6Text     sex     
##  Min.   :   1   Keep  :1613   GB     :1560   Age 18-24:365   Female:727  
##  1st Qu.: 409   Remove:  20   FR     :  10   Age 25-34:232   Male  :906  
##  Median : 817                 US     :  10   Age 35-44:210               
##  Mean   : 817                 ES     :   5   Age 45-54:293               
##  3rd Qu.:1225                 GR     :   5   Age 55-64:301               
##  Max.   :1633                 DE     :   3   Age 65+  :232               
##                               (Other):  40                               
##   previousVote intendReferendumVote ethnicity    religion           Q1       
##  Labour :557   Leave :690           a:1541    Min.   :1.000   Min.   :1.000  
##  UKIP   :305   NoVote: 22           b:  36    1st Qu.:1.000   1st Qu.:2.000  
##  Con    :250   Remain:907           c:  26    Median :2.000   Median :3.000  
##  LibDem :182   Spoil : 14           d:   8    Mean   :2.333   Mean   :3.476  
##  Green  :164                        e:  22    3rd Qu.:4.000   3rd Qu.:5.000  
##  NoVote : 86                                  Max.   :5.000   Max.   :9.000  
##  (Other): 89                                                                 
##        Q2              Q3              Q4              Q5       
##  Min.   :1.000   Min.   :1.000   Min.   :1.000   Min.   :1.000  
##  1st Qu.:1.000   1st Qu.:5.000   1st Qu.:1.000   1st Qu.:1.000  
##  Median :1.000   Median :8.000   Median :1.000   Median :3.000  
##  Mean   :1.777   Mean   :6.921   Mean   :2.078   Mean   :3.132  
##  3rd Qu.:1.000   3rd Qu.:9.000   3rd Qu.:2.000   3rd Qu.:5.000  
##  Max.   :9.000   Max.   :9.000   Max.   :9.000   Max.   :9.000  
##                                                                 
##        Q6              Q7              Q8              Q9       
##  Min.   :1.000   Min.   :1.000   Min.   :1.000   Min.   :1.000  
##  1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000  
##  Median :1.000   Median :3.000   Median :2.000   Median :2.000  
##  Mean   :2.277   Mean   :3.628   Mean   :2.691   Mean   :2.484  
##  3rd Qu.:3.000   3rd Qu.:6.000   3rd Qu.:4.000   3rd Qu.:3.000  
##  Max.   :9.000   Max.   :9.000   Max.   :9.000   Max.   :9.000  
##                                                                 
##       Q10             Q11             Q12             Q13       
##  Min.   :1.000   Min.   :1.000   Min.   :1.000   Min.   :1.000  
##  1st Qu.:1.000   1st Qu.:1.000   1st Qu.:2.000   1st Qu.:1.000  
##  Median :3.000   Median :1.000   Median :4.000   Median :2.000  
##  Mean   :3.851   Mean   :2.134   Mean   :4.366   Mean   :2.798  
##  3rd Qu.:6.000   3rd Qu.:3.000   3rd Qu.:6.000   3rd Qu.:4.000  
##  Max.   :9.000   Max.   :9.000   Max.   :9.000   Max.   :9.000  
##                                                                 
##       Q14             Q15             Q16             Q17       
##  Min.   :1.000   Min.   :1.000   Min.   :1.000   Min.   :1.000  
##  1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000  
##  Median :3.000   Median :3.000   Median :1.000   Median :3.000  
##  Mean   :4.089   Mean   :3.119   Mean   :2.018   Mean   :3.727  
##  3rd Qu.:6.000   3rd Qu.:4.000   3rd Qu.:2.000   3rd Qu.:6.000  
##  Max.   :9.000   Max.   :9.000   Max.   :9.000   Max.   :9.000  
##                                                                 
##       Q18             Q19             Q20             Q21       
##  Min.   :1.000   Min.   :1.000   Min.   :1.000   Min.   :1.000  
##  1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000  
##  Median :1.000   Median :2.000   Median :2.000   Median :3.000  
##  Mean   :1.644   Mean   :3.158   Mean   :2.266   Mean   :3.386  
##  3rd Qu.:1.000   3rd Qu.:5.000   3rd Qu.:3.000   3rd Qu.:5.000  
##  Max.   :9.000   Max.   :9.000   Max.   :9.000   Max.   :9.000  
##                                                                 
##       Q22       
##  Min.   :1.000  
##  1st Qu.:1.000  
##  Median :2.000  
##  Mean   :3.299  
##  3rd Qu.:5.000  
##  Max.   :9.000  
##

2.2 Remove unknowns/outliers

“Remove” column indicates the data points which shouldn’t be considered. Those records were recorded within a minute, which is not possible for this survey.
Removing “NoVote” and “Spoil” from “intendReferendumVote” as we are only interested in how they vote.
Reverse coding is not needed for this dataset, it’s recorded in the correct order. (Altemeyer 2006)
All the records were in the scale, no check for outliers was necessary.

data = subset(master, master$Remove == "Keep")
data = subset(data, data$intendReferendumVote != "Spoil")
data = subset(data, data$intendReferendumVote != "NoVote")
data = droplevels(data)
data$Vote = as.numeric(data$intendReferendumVote)-1
summary(data)

##        id          Remove        country       ageGroup6Text     sex     
##  Min.   :   1.0   Keep:1577   GB     :1505   Age 18-24:352   Female:703  
##  1st Qu.: 413.0               FR     :  10   Age 25-34:227   Male  :874  
##  Median : 817.0               US     :  10   Age 35-44:200               
##  Mean   : 817.5               ES     :   5   Age 45-54:279               
##  3rd Qu.:1225.0               GR     :   4   Age 55-64:294               
##  Max.   :1633.0               DE     :   3   Age 65+  :225               
##                               (Other):  40                               
##   previousVote intendReferendumVote ethnicity    religion           Q1       
##  Labour :538   Leave :681           a:1491    Min.   :1.000   Min.   :1.000  
##  UKIP   :298   Remain:896           b:  34    1st Qu.:1.000   1st Qu.:2.000  
##  Con    :242                        c:  24    Median :2.000   Median :3.000  
##  LibDem :179                        d:   7    Mean   :2.325   Mean   :3.489  
##  Green  :157                        e:  21    3rd Qu.:4.000   3rd Qu.:5.000  
##  NoVote : 76                                  Max.   :5.000   Max.   :9.000  
##  (Other): 87                                                                 
##        Q2             Q3              Q4              Q5              Q6       
##  Min.   :1.00   Min.   :1.000   Min.   :1.000   Min.   :1.000   Min.   :1.000  
##  1st Qu.:1.00   1st Qu.:5.000   1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000  
##  Median :1.00   Median :9.000   Median :1.000   Median :3.000   Median :1.000  
##  Mean   :1.76   Mean   :6.937   Mean   :2.064   Mean   :3.124   Mean   :2.261  
##  3rd Qu.:1.00   3rd Qu.:9.000   3rd Qu.:2.000   3rd Qu.:5.000   3rd Qu.:3.000  
##  Max.   :9.00   Max.   :9.000   Max.   :9.000   Max.   :9.000   Max.   :9.000  
##                                                                                
##        Q7              Q8              Q9             Q10       
##  Min.   :1.000   Min.   :1.000   Min.   :1.000   Min.   :1.000  
##  1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000  
##  Median :3.000   Median :2.000   Median :2.000   Median :3.000  
##  Mean   :3.635   Mean   :2.679   Mean   :2.469   Mean   :3.856  
##  3rd Qu.:6.000   3rd Qu.:4.000   3rd Qu.:3.000   3rd Qu.:6.000  
##  Max.   :9.000   Max.   :9.000   Max.   :9.000   Max.   :9.000  
##                                                                 
##       Q11             Q12             Q13             Q14       
##  Min.   :1.000   Min.   :1.000   Min.   :1.000   Min.   :1.000  
##  1st Qu.:1.000   1st Qu.:2.000   1st Qu.:1.000   1st Qu.:1.000  
##  Median :1.000   Median :4.000   Median :2.000   Median :4.000  
##  Mean   :2.136   Mean   :4.375   Mean   :2.788   Mean   :4.098  
##  3rd Qu.:3.000   3rd Qu.:6.000   3rd Qu.:4.000   3rd Qu.:6.000  
##  Max.   :9.000   Max.   :9.000   Max.   :9.000   Max.   :9.000  
##                                                                 
##       Q15             Q16         Q17             Q18             Q19      
##  Min.   :1.000   Min.   :1   Min.   :1.000   Min.   :1.000   Min.   :1.00  
##  1st Qu.:1.000   1st Qu.:1   1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.00  
##  Median :3.000   Median :1   Median :3.000   Median :1.000   Median :2.00  
##  Mean   :3.113   Mean   :2   Mean   :3.706   Mean   :1.628   Mean   :3.15  
##  3rd Qu.:4.000   3rd Qu.:2   3rd Qu.:6.000   3rd Qu.:1.000   3rd Qu.:5.00  
##  Max.   :9.000   Max.   :9   Max.   :9.000   Max.   :9.000   Max.   :9.00  
##                                                                            
##       Q20             Q21             Q22             Vote       
##  Min.   :1.000   Min.   :1.000   Min.   :1.000   Min.   :0.0000  
##  1st Qu.:1.000   1st Qu.:1.000   1st Qu.:1.000   1st Qu.:0.0000  
##  Median :2.000   Median :3.000   Median :2.000   Median :1.0000  
##  Mean   :2.255   Mean   :3.384   Mean   :3.292   Mean   :0.5682  
##  3rd Qu.:3.000   3rd Qu.:5.000   3rd Qu.:5.000   3rd Qu.:1.0000  
##  Max.   :9.000   Max.   :9.000   Max.   :9.000   Max.   :1.0000  
##

3 Overall Model

For the theory we want to test, all the question are tied to a variable and that variable will predict the voting preference.

4 Sex Based Model

Checking to see if Gender has any role on the voting preference.

4.1 Prepare Data for Gender Based Model

Building a one-factor model for this analysis.
There are 2 factors in this model, Male and Female.

4.1.1 Female Model

gender.fit.female = cfa(model = brexit.model, data = data[data$sex == "Female",], meanstructure = T)

4.1.2 Male Model

gender.fit.male = cfa(model = brexit.model, data = data[data$sex == "Male",], meanstructure = T)

4.1.3 Gender Configural Model

gender.configural.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "sex")

4.1.4 Gender Metric Model

gender.metric.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "sex", group.equal = c("loadings"))

4.1.5 Gender Scalar Model

gender.scalar.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "sex", group.equal = c("loadings", "intercepts"))

4.1.6 Gender Strict Model

gender.strict.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "sex", group.equal = c("loadings", "intercepts", "residuals"))

4.1.7 Gender Partial Invariance Model

gender_partial_syntax = paste(colnames(data)[10:31], "~~", colnames(data)[10:31])
gender_CFI_list  <- 1:length(gender_partial_syntax)
names(gender_CFI_list) <- gender_partial_syntax

for (i in 1:length(gender_partial_syntax)){
  temp <- cfa(model = brexit.model, 
              data = data,
              meanstructure = TRUE,
              group = "sex",
              group.equal = c("loadings", "intercepts", "residuals"),
              group.partial = gender_partial_syntax[i])
  gender_CFI_list[i] <- fitmeasures(temp, "cfi")
}

gender.partial.fit <- cfa(model = brexit.model, data = data, meanstructure = TRUE, group = "sex", group.equal = c("loadings", "intercepts", "residuals"), group.partial = names(which.max(gender_CFI_list)))

4.1.8 Gender Output Matrix

None of them are great predictors of voting patterns, it can be safely concluded that gender doesn’t play a role in any apparent pattern in Brexit Referendum Voting preference.
The model for Male was slightly skewed, but not significant enough to determine anything.

Gender CFA Model
Model	X2	df	CFI	RMSEA	SRMR
Overall Model	5542.804	230	0.698	0.121	0.108
Female Gender Model	3210.014	230	0.635	0.136	0.127
Male Gender Model	2817.605	230	0.733	0.113	0.097
Gender Configural Model	6027.62	460	0.688	0.124	0.11
Gender Metric Model	6082.895	481	0.686	0.122	0.113
Gender Scalar Model	6264.688	503	0.678	0.121	0.114
Gender Strict Model	6391.618	526	0.672	0.119	0.116
Gender Partial Invariance Model	6361.703	525	0.673	0.119	0.115

5 Religion Based Model

Checking to see if religion plays a role in voting preference.

5.1 Preparing Data for Religion

There are 5 factors on Religion Column. They are only named as 1, 2, 3, 4 and 5.
It was not necessary to figure out the religion names as it doesn’t concern the model.

5.1.1 Religion(1) Model

rel.one.model = cfa(model = brexit.model, data = data[data$religion == 1,], meanstructure = T)

5.1.2 Religion(2) Model

rel.two.model = cfa(model = brexit.model, data = data[data$religion == 2,], meanstructure = T)

5.1.3 Religion(3) Model

rel.three.model = cfa(model = brexit.model, data = data[data$religion == 3,], meanstructure = T)

5.1.4 Religion(4) Model

rel.four.model = cfa(model = brexit.model, data = data[data$religion == 4,], meanstructure = T)

5.1.5 Religion(5) Model

rel.five.model = cfa(model = brexit.model, data = data[data$religion == 5,], meanstructure = T)

5.1.6 Religion Configural Model

rel.configural.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "religion")

5.1.7 Religion Metric Model

rel.metric.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "religion", group.equal = c("loadings"))

5.1.8 Religion Scalar Model

rel.scalar.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "religion", group.equal = c("loadings", "intercepts"))

5.1.9 Religion Strict Model

rel.strict.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "religion", group.equal = c("loadings", "intercepts", "residuals"))

5.1.10 Religion Partial Invariance Model

rel_partial_syntax = paste(colnames(data)[10:31], "~~", colnames(data)[10:31])
rel_CFI_list  <- 1:length(rel_partial_syntax)
names(rel_CFI_list) <- rel_partial_syntax

for (i in 1:length(rel_partial_syntax)){
  temp <- cfa(model = brexit.model, 
              data = data,
              meanstructure = TRUE,
              group = "religion",
              group.equal = c("loadings", "intercepts", "residuals"),
              group.partial = rel_partial_syntax[i])
  rel_CFI_list[i] <- fitmeasures(temp, "cfi")
}

rel.partial.fit <- cfa(model = brexit.model, data = data, meanstructure = TRUE, group = "religion", group.equal = c("loadings", "intercepts", "residuals"), group.partial = names(which.max(rel_CFI_list)))

5.1.11 Religion Output Matrix

None of them are great predictors of voting patterns, it can be safely concluded that Religion doesn’t play a role in any apparent pattern in Brexit Referendum Vote
Results from Religion’s 1 and 2 are slightly monotonous but agagin, nothing conclusive.

Religion CFA Model
Model	X2	df	CFI	RMSEA	SRMR
Overall Model	5542.804	230	0.698	0.121	0.108
Religion(1) Model	1971.328	230	0.733	0.1	0.089
Religion(2) Model	674.461	230	0.712	0.111	0.086
Religion(3) Model	1224.211	230	0.597	0.134	0.148
Religion(4) Model	1236.903	230	0.665	0.129	0.123
Religion(5) Model	1167.258	230	0.586	0.157	0.144
Religion Configural Model	6274.161	1150	0.676	0.119	0.109
Religion Metric Model	6515.192	1234	0.666	0.116	0.121
Religion Scalar Model	7120.566	1322	0.633	0.118	0.123
Religion Strict Model	8382.156	1414	0.559	0.125	0.151
Religion Partial Invariance Model	8175.329	1410	0.572	0.123	0.141

6 Age Group Based Model

Checking to see if age groups play a role in voting preference.

6.1 Preparing Data for Age Group Analysis

There are 6 factors in AgeGroup; Age 18-24, Age 25-34, Age 35-44, Age 45-54, Age 55-64, and Age 65+

6.1.1 Age Group “Age 18-24” Model

age.one.fit = cfa(model = brexit.model, data = data[data$ageGroup6Text == "Age 18-24",], meanstructure = T)

6.1.2 Age Group “Age 25-34” Model

age.two.fit = cfa(model = brexit.model, data = data[data$ageGroup6Text == "Age 25-34",], meanstructure = T)

6.1.3 Age Group “Age 35-44” Model

age.three.fit = cfa(model = brexit.model, data = data[data$ageGroup6Text == "Age 35-44",], meanstructure = T)

6.1.4 Age Group “Age 45-54” Model

age.four.fit = cfa(model = brexit.model, data = data[data$ageGroup6Text == "Age 45-54",], meanstructure = T)

6.1.5 Age Group “Age 55-64” Model

age.five.fit = cfa(model = brexit.model, data = data[data$ageGroup6Text == "Age 55-64",], meanstructure = T)

6.1.6 Age Group “Age 65+” Model

age.six.fit = cfa(model = brexit.model, data = data[data$ageGroup6Text == "Age 65+",], meanstructure = T)

6.1.7 Age Group Configural Model

age.configural.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "ageGroup6Text")

6.1.8 Age Group Metric Model

age.metric.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "ageGroup6Text", group.equal = c("loadings"))

6.1.9 Age Group Scalar Model

age.scalar.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "ageGroup6Text", group.equal = c("loadings", "intercepts"))

6.1.10 Age Group Strict Model

age.strict.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "ageGroup6Text", group.equal = c("loadings", "intercepts", "residuals"))

6.1.11 Age Group Partial Invariance Model

age_partial_syntax = paste(colnames(data)[10:31], "~~", colnames(data)[10:31])
age_CFI_list  <- 1:length(age_partial_syntax)
names(age_CFI_list) <- age_partial_syntax

for (i in 1:length(age_partial_syntax)){
  temp <- cfa(model = brexit.model, 
              data = data,
              meanstructure = TRUE,
              group = "ageGroup6Text",
              group.equal = c("loadings", "intercepts", "residuals"),
              group.partial = age_partial_syntax[i])
  age_CFI_list[i] <- fitmeasures(temp, "cfi")
}

age.partial.fit <- cfa(model = brexit.model, data = data, meanstructure = TRUE, group = "ageGroup6Text", group.equal = c("loadings", "intercepts", "residuals"), group.partial = names(which.max(age_CFI_list)))

6.1.12 Age Group Output Matrix

None of them are great predictors of voting patterns, it can be safely concluded that Age Group doesn’t play a role in any apparent pattern in Brexit Referendum Vote.
Similar results across the board. Nothing significant to conclude our theory of voting preference being informed by age.

Age Group CFA Model
Model	X2	df	CFI	RMSEA	SRMR
Overall Model	5542.804	230	0.698	0.121	0.108
Age Group (18-24) Model	1355.441	230	0.757	0.118	0.081
Age Group (25-34) Model	926.648	230	0.662	0.116	0.099
Age Group (35-44) Model	940.729	230	0.725	0.124	0.1
Age Group (45-54) Model	1169.411	230	0.682	0.121	0.102
Age Group (55-64) Model	1494.714	230	0.597	0.137	0.15
Age Group (65+) Model	1344.977	230	0.556	0.147	0.171
Age Group Configural Model	7231.92	1380	0.673	0.127	0.115
Age Group Metric Model	7619.473	1485	0.657	0.125	0.13
Age Group Scalar Model	8070.001	1595	0.638	0.124	0.134
Age Group Strict Model	9675.02	1710	0.554	0.133	0.153
Age Group Partial Invariance Model	9460.732	1705	0.566	0.132	0.15

7 Previous Vote Preference Based Model

7.1 Preparing Data for Previous Vote Analysis

There are 8 factors in PreviousVote column: Con, Cymru, Green, Labour, LibDem, NoVote, SNP, and UKIP
Dropping CYMRU as there are not enough data points for the analysis

7.1.1 UKIP Model

data = subset(data, data$previousVote != "Cymru")
data = droplevels(data)
UKIP.fit = cfa(model = brexit.model, data = data[data$previousVote == "UKIP",], meanstructure = T)

7.1.2 SNP Model

SNP.fit = cfa(model = brexit.model, data = data[data$previousVote == "SNP",], meanstructure = T)

7.1.3 NoVote Model

NoVote.fit = cfa(model = brexit.model, data = data[data$previousVote == "NoVote",], meanstructure = T)

7.1.4 LibDem Model

LibDem.fit = cfa(model = brexit.model, data = data[data$previousVote == "LibDem",], meanstructure = T)

7.1.5 Labour Model

Labour.fit = cfa(model = brexit.model, data = data[data$previousVote == "Labour",], meanstructure = T)

7.1.6 Green Model

Green.fit = cfa(model = brexit.model, data = data[data$previousVote == "Green",], meanstructure = T)

7.1.7 Con Model

Con.fit = cfa(model = brexit.model, data = data[data$previousVote == "Con",], meanstructure = T)

7.1.8 PreviousVote Configural Model

pv.configural.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "previousVote")

7.1.9 PreviousVote Metric Model

pv.metric.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "previousVote", group.equal = c("loadings"))

7.1.10 PreviousVote Scalar Model

pv.scalar.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "previousVote", group.equal = c("loadings", "intercepts"))

7.1.11 PreviousVote Strict Model

pv.strict.fit = cfa(model = brexit.model, data = data, meanstructure = T, group = "previousVote", group.equal = c("loadings", "intercepts", "residuals"))

7.1.12 previous Vote Invariance Model

pv_partial_syntax = paste(colnames(data)[10:31], "~~", colnames(data)[10:31])
pv_CFI_list  <- 1:length(pv_partial_syntax)
names(pv_CFI_list) <- pv_partial_syntax

for (i in 1:length(pv_partial_syntax)){
  temp <- cfa(model = brexit.model, 
              data = data,
              meanstructure = TRUE,
              group = "previousVote",
              group.equal = c("loadings", "intercepts", "residuals"),
              group.partial = pv_partial_syntax[i])
  pv_CFI_list[i] <- fitmeasures(temp, "cfi")
}

pv.partial.fit <- cfa(model = brexit.model, data = data, meanstructure = TRUE, group = "previousVote", group.equal = c("loadings", "intercepts", "residuals"), group.partial = names(which.max(pv_CFI_list)))

7.1.13 Previous Vote Output Matrix

None of them are great predictors of voting patterns, it can be safely concluded that Previous Voting Preference doesn’t play a role in any apparent pattern in Brexit Referendum Vote.
Previous Voting preference is not a good indicator of Brexit Voting preference.

Previous Vote Preference CFA Model
Model	X2	df	CFI	RMSEA	SRMR
Overall Model	5542.804	230	0.698	0.121	0.108
UKIP Model	1714.059	230	0.441	0.147	0.163
SNP Model	496.966	230	0.524	0.124	0.127
NoVote Model	489.806	230	0.612	0.122	0.118
LibDem Model	726.646	230	0.647	0.11	0.099
Labour Model	1319.282	230	0.741	0.094	0.082
Green Model	697.766	230	0.735	0.114	0.083
Con Model	1225.77	230	0.582	0.134	0.136
Previous Vote Configural Model	6670.295	1610	0.629	0.119	0.112
Previous Vote Metric Model	7221.5	1736	0.598	0.119	0.13
Previous Vote Scalar Model	8636.344	1868	0.504	0.127	0.241
Previous Vote Strict Model	11183.985	2006	0.327	0.143	0.377
Previous Vote Partial Model	10899.608	2000	0.348	0.141	0.373

8 Discussion

The brexit referendum vote was almost a 50-50 vote. This model somewhat replicates that. We don’t see a clear division. The results are all over the place. The fit measure scores are bad but not terrible, but they aren’t good either.
As is the case with all analysis, more data could have normalized or spread out the variable to get maybe more accurate results.

References

Altemeyer, Bob. 2006. “Chapter 1.” In The Authoritarians. University of Manitoba. https://web.archive.org/web/20160824064748/http://members.shaw.ca/jeanaltemeyer/drbob/chapter1.pdf.

Altemeyer, Bob, and Chris W Johnson. 2020. The RWA Scale. http://www.panojohnson.com/automatons/rwa-scale.xhtml.

Ghosh, Iman. 2019. “Charts: America’s Political Divide, 1994–2017.” Visual Capitalist. https://www.visualcapitalist.com/charts-americas-political-divide-1994-2017/.

Thomas Carothers, Andrew O’Donohue. 2019. “How to Understand the Global Spread of Political Polarization.” Carnegie Endowment for International Peace. https://carnegieendowment.org/2019/10/01/how-to-understand-global-spread-of-political-polarization-pub-79893.

Final Project

Rushabh Barbhaya

2/15/2021